Contribution to the Biennale Architettura 2025
Ramon Elias Weber, PhD, Assistant Professor UC Berkeley
How can we create the next generation of architectural design tools to design more sustainable buildings? How can we invent new data structures and methods of representation that are specific to architecture? How can we integrate artificial intelligence into design processes? How can we augment human intuition with machine intelligence to design better buildings?
The project presents a physical snapshot of a digital algorithm to represent, generate and manipulate building floor plans. The new generative procedure and its use in estimating and reducing carbon emissions have been shown in the scientific publication in Nature Communications.
Background and Research
The published research is available open access: Nature Communications 2024. A hypergraph model shows the carbon reduction potential of effective space use in housing. Ramon Elias Weber, Caitlin Mueller, Christoph Reinhart
The new algorithm for the generative design of building floor plans – the hypergraph – enables the real-time creation and manipulation of the complex geometric relationships found in architectural floor plans.
The hypergraph captures both the spatial subdivisions and adjacency relationships between the different rooms. With this, each floor plan has a specific and unique hypergraph that represents its spatial topology.
The same hypergraph can be applied to different boundaries to create floor plan. Different hypergraphs can be applied to the same boundary to create different internal configurations.
New design tools for architects allow real-time manipulation of floor plans. Integrating directly with building performance simulation tools, automated building energy, daylight and embodied carbon analysis can be conducted.
How can we best learn from the design intelligence of existing buildings? How can we best integrate new types of Artificial Intelligence into the design of new buildings?
The Installation
The installation at the 19th International Architecture Exhibition of La Biennale di Venezia, curated by Carlo Ratti, creates a physical snapshot of the digital floor plan automation algorithms.
A database of 144 floor plans is distributed over a plinth. Each unique floor plan has a corresponding hypergraph representation. The nodes and lines in the hypergraphs are physically represented through 3D printed nodes and aluminum struts.
Four buildings with the same outlines hover over the database and pick and choose different floorplans underneath them. The hypergraphs of the chosen floorplans are extracted and projected into the building. The selected floorplans are adjusted to the new apartment boundaries inside the building and create different internal configurations. An environmental simulation assesses the performance of each layout in terms of daylight, energy use and spatial configuration.
Digital to Physical
Implications for Architectural Design
The algorithms create opportunities for new types of workflows in architectural design. Starting from a library of existing designs, floor plans represented with the hypergraph can be adjusted, generated automatically, or used to populate a building boundary.
How could we rethink our cities, leveraging new technologies, artificial intelligence and with them, infinite design possibilities?
The work was supported by the University of California, Berkeley, and Cass Calder Smith Architects.
Team:
Ramon Elias Weber, PhD, Assistant Professor UC Berkeley
Yuhan Zhang, Frederic Lam, Zhuoer Chen, Leah Altman, Linc Ruiz-Truong, Mia Wilson, Emma Nakoaka, Sihyeok Yang and Noel Fäh.
Please contact ramon@berkeley.edu for more information. Download images in high resolution here.